The US and China Are Collaborating More Closely on AI Than You Think
The United States and China is, in many ways, a rival in the field of artificial intelligence, with companies racing to outdo each other in algorithms, models and specialized silicon. And yet, the world’s AI superpowers still collaborate to a surprising degree when it comes to cutting-edge research.
A WIRED analysis of more than 5,000 AI research papers presented last month at the industry’s leading conference, Neural Information Processing Systems (NeurIPS), reveals significant collaboration between U.S. and Chinese labs.
The analysis found that 141 of the total 5,290 articles (about 3%) involve collaboration between authors affiliated with U.S. institutions and those affiliated with Chinese institutions. Collaboration between the United States and China also appears fairly consistent, with 134 papers out of 4,497 total involving authors from institutions in both countries in 2024.
WIRED also studied how algorithms and models developed in one country are shared and adapted across the Pacific. The transformer architecture, developed by a team of Google researchers and now widely used in industry, is presented in 292 papers by authors from Chinese institutions. Meta’s Llama family of models was a key part of the research presented in 106 of these papers. Meanwhile, Chinese tech giant Alibaba’s increasingly popular Qwen language model appears in 63 articles including authors from U.S. organizations.
Jeffrey Ding, an assistant professor at George Washington University who follows China’s AI landscape, says he’s not surprised to see such a level of teamwork. “Whether policymakers on both sides like it or not, the US and Chinese AI ecosystems are inextricably linked – and both benefit from this agreement,” says Ding.
The analysis undoubtedly simplifies the extent to which the United States and China share ideas and talents. Many researchers of Chinese origin study in the United States and form bonds with their colleagues that last a lifetime.
“NeurIPS itself is an example of international collaboration and a testament to its importance to our field,” Katherine Gorman, a spokesperson for NeurIPS, said in a statement. “Collaborations between students and advisors often continue long after the student has left their university. You can see these kinds of signals that indicate cooperation across the board in many places, including in professional networks and former collaborators.”
The latest issue of WIRED explores the many ways China is shaping the current century. But as U.S. politicians and tech executives use fears over China’s rise as justification to abandon regulations and fuel sky-high investments, our analysis is a good reminder that the world’s two AI superpowers still have much to gain by working together.
A note on methodology
I used Codex, OpenAI’s model for writing code, to help analyze the NeurIPS articles. After writing a script to download all the articles, I used the template to explore each one and do some analysis. This involved Codex writing a script to search for American and Chinese institutions in the author field of each article.
The experiment offered a fascinating insight into the potential of coding models to automate useful tasks. There’s a lot of panic about AI replacing coding tasks, but it’s something I normally wouldn’t have had the time or budget to build. I started by writing scripts and asking Codex to modify them before simply asking Codex to do the analysis itself. This involved importing Python libraries, testing different tools and writing scripts before producing reports for review. The process involved quite a bit of trial and error, and you have to be very careful, because AI models make surprisingly stupid mistakes, even when they’re quite intelligent. I had to make sure each report included a way to browse the results, and I checked as many of them manually as possible.
This is an edition of Will Knight’s AI Lab Newsletter. Read previous newsletters here.
